Introduction and parameters
At first glance, the Gigabyte GeForce RTX 5080 Gaming OC isn’t striking in terms of flashy design or vibrant lighting. It’s neither the cheapest nor the most expensive model. In a standard setup, the lighting is subtle, but the backlit fans stand out more when mounted vertically. The real advantage, though, is its oversized cooler and well-tuned fan control, allowing it to outperform even pricier competitors in terms of overall performance.
A detailed breakdown of the new technologies introduced with these GPUs and their architecture has been written by Jan Olšan in the article Blackwell: GeForce RTX 5000 architecture and innovations. This article covers the individual chips used in Nvidia’s latest graphics cards, the introduction of GDDR7 memory, cache improvements, updates to GPU compute units, changes to shaders, tensor core optimizations, and improved ray tracing units. The new cards also feature support for DisplayPort 2.1b and PCI Express 5.0.
With the new generation of cards, the capabilities of DLSS are also advancing, which can now use Multi Frame Generation to generate up to three intermediate frames, and new is also the second generation of Nvidia Reflex with Frame Warp technology, which can adjust the movement of objects and the camera in rendered frames based on the most recent information from peripherals, which it receives only during or after the frame is rendered. Everything is described in more detail in the mentioned article.
We’ve already discussed the reference specifications, so I’ll just summarize them in a table and move on to the card itself, which benefits from factory overclocking and slightly higher clock speeds.
GeForce RTX 5090 | GeForce RTX 5080 | GeForce RTX 4090 | GeForce RTX 4080 Super | GeForce RTX 4080 | |
Shader cores | Blackwell | Blackwell | Ada Lovelace | Ada Lovelace | Ada Lovelace |
Nvidia CUDA cores (SP) | 21760 | 10752 | 16384 | 10240 | 9728 |
Texture Mapping Units | 680 | 336 | 512 | 320 | 304 |
Render Output Units | 176 | 128 | 176 | 112 | 112 |
RT cores | 318 TFLOPS | 171 TFLOPS | 191 TFLOPS | 121 TFLOPS | 113 TFLOPS |
Tensor Cores | 3352 AI TOPS | 1801 AI TOPS | 1321 AI TOPS | 836 AI TOPS | 780 AI TOPS |
Boost Clock | 2.41 GHz | 2.62 GHz | 2.52 GHz | 2.55 GHz | 2.51 GHz |
Base Clock | 2.10 GHz | 2.30 GHz | 2.23 GHz | 2.29 GHz | 2.21 GHz |
Memory | 32 GB GDDR7 | 16 GB DDR7 | 24 GB GDDR6X | 16 GB GDDR6X | 16 GB GDDR6X |
Memory Bus Width | 512-bit | 256-bit | 384-bit | 256-bit | 256-bit |
Memory Bandwidth | 1792 GB/s | 960 GB/s | 1008 GB/s | 736 GB/s | 717 GB/s |
Max. resolution | 4K at 480Hz or 8K at 120Hz with DSC | 4K at 480Hz or 8K at 120Hz with DSC | 4K at 240 Hz or 8K at 60 Hz with DSC, HDR | 4K at 240 Hz or 8K at 60 Hz with DSC, HDR | 4K at 240 Hz or 8K at 60 Hz with DSC, HDR |
Standard Display Connectors | 3× DisplayPort, 1× HDMI | 3× DisplayPort, 1× HDMI | 3× DisplayPort, 1× HDMI | 3× DisplayPort, 1× HDMI | 3× DisplayPort, 1× HDMI |
PCI Express | Gen 5 | Gen 5 | Gen 4 | Gen 4 | Gen 4 |
Nvidia Encoder | 3× 9th gen | 2× 9th gen | 2× 8th gen | 2× 8th gen | 2× 8th gen |
Nvidia Decoder | 2× 6th gen | 2× 6th gen | 1× 5th gen | 1× 5th gen | 1× 5th gen |
length | 304 mm | 304 mm | 304 mm | 304 mm | 304 mm |
width | 137 mm | 137 mm | 137 mm | 137 mm | 137 mm |
height | 2-slot | 2-slot | 3-slot | 3-slot | 3-slot |
Total Graphics Power | 575 W | 360 W | 450 W | 320 W | 320 W |
Supplementary Power Connectores | 4× 8pin PCIe (adapter) or 1× 600 W PCIe Gen 5 | 3× 8pin PCIe (adapter) or 1× 450 W PCIe Gen 5 | 3× 8pin PCIe (adapter) or 1× 450 W PCIe Gen 5 | 3× 8pin PCIe (adapter) or 1× 450 W PCIe Gen 5 | 3× 8pin PCIe (adapter) or 1× 450 W PCIe Gen 5 |
The next GeForce RTX 5080 we will test is a factory-overclocked model from Gigabyte’s Gaming series. Compared to the ROG Astral RTX 5080 OC Edition, which we reviewed previously, the RTX 5080 Gaming OC offers a lower price, more standard dimensions, a slightly sleeker design, and slightly lower performance. It sits roughly in the middle of Gigabyte’s lineup, between the more affordable Windforce series and the high-end Aorus Master and Aorus Xtreme models.
The listed factory boost clock speed for the chip has been increased from the reference 2617 MHz to 2730 MHz. However, as is often the case, real-world results tend to be slightly better. The actual operating clock speeds are dynamically regulated based on the card’s workload, power consumption, and chip temperatures. Detailed graphs illustrating these clock speed trends can be found in the performance testing sections.
The new version of GPU-Z 2.62.0 is now available for download on TechPowerUp and is fully compatible with the GeForce RTX 50 series. The screenshots show the Quiet BIOS, which has a power limit set at 360 W, with the option to increase it to 400 W. The Performance BIOS, on the other hand, has a fixed upper limit of 400 W, which cannot be increased further. Additionally, the two BIOS modes differ in their fan speed curve settings–Performance mode runs at higher fan speeds. However, this information cannot be directly read from GPU-Z.
The second BIOS is Performance. The power limit is the same, as well as the extent to which it can be increased.
To the monitoring values displayed by GPU-Z, I will also add a list of all sensors and trackable parameters from the latest beta version of HWiNFO, which now also includes support for the RTX 50 series.
⠀
- Contents
- Introduction and parameters
- Card design, connectors
- Test build
- Black Myth: Wukong, Metro Exodus Enhanced Edition
- Cyberpunk 2077
- F1 24
- 3DMark Port Royal, Blender
- Operating characteristics – Cyberpunk, BIOS quiet
- Operating characteristics – Cyberpunk, BIOS quiet
- Operating characteristics – F1 24, performance
- Operating characteristics – F1 24, quiet
- F1 24, maximum power limit + overclocking
- Closing summary